In Defence of `This Worldly' Causality: Comments on van Fraassen's Laws and Symmetry

Abstract

Recall van Fraassen's general scheme. Scientific models can depict both observable and unobservable processes. Science itself is supposed to tell us what is and what is not observable. We are to accept the theory if it gets right all that is observable; the rest can be of no matter to us, van Fraassen has suggested, because we could have no evidence about whether the model gets it right or not. We may believe in the model's picture of the unobservable world but that belief is based on faith not evidence, and should be recognised as such. Most of the stuff of modem physics, then, turns out to be in the unobservable part of the model: photons, quarks, gravity-waves, the electromagnetic field, forces, I suppose and even space-time itself. Where are causes? that is the question I want to discuss. We can begin with causal laws. They are surely for van Fraassen not in the model itself any more than any other kinds of laws. They enter, rather, as a constraint on the kinds of structures we will allow our models to have. In a full model for the entire history of the world, or more realistically, in a small model for a given scientific experiment, event-kinds will recur with some frequency. These frequencies are represented by probabilities-though note that the probabilities are not in the models either. If we insist that our scientific theory in a given domain be causal, that will restrict the form of the probability functions which are allowed and hence, in some way, constrain the actual models, which have finite frequencies in them only. Some patterns of correlation will be allowed and some will be prohibited. Consider van Fraassen's discussion of the Einstein-Podalsky-Rosen (EPR) paradox in the follow-up to Laws and Symmetry, Quantum Mechanics: An Empiricist View.' There van Fraassen takes a particular relationship between probabilities-calledfactoriz-